On the Generalization Capability of Multi-Layered Networks in the Extraction of Speech Properties

The paper describes a speech coding system
based on an ear model followed by a set of Multi-
Layer Networks (MLN). MLNs are trained to learn
how to recognize articulatory features like the
place and manner of articulation. Experiments are
performed on 10 English vowels showing a
recognition rate higher than 95% for new
speakers. When features are used for recognition,
comparable results are obtained for vowels and
diphthongs not used for training and pronounced
by new speakers. This suggests that MLNs
suitably fed by the data computed by an ear model
have good generalization capabilities over new
speakers and new sounds.

Tipo Pubblicazione: 
Contributo in atti di convegno
Author or Creator: 
De Mori R.
Bengio Y.
Cosi P.
Publisher: 
Morgan Kaufmann Publishers Inc., San Francisco, CA, USA
Source: 
IEEE IJCAI-89 - International Joint Conference on Artificial Intelligence, pp. 1531–1536, Detroit, MI, U.S.A., August 20-25, 1989
Date: 
1989
Resource Identifier: 
http://www.cnr.it/prodotto/i/241918
https://dx.doi.org/10.1.1.76.2662
info:doi:10.1.1.76.2662
http://www.ijcai.org/Past%20Proceedings/IJCAI-89-VOL-2/PDF/108.pdf
Language: 
Eng
ISTC Author: 
Ritratto di Piero Cosi
Real name: